Re: I-D Action: draft-mishra-6man-variable-slaac-01.txt

[Alexandre Petrescu <[e-mail]>]

Le 03/11/2020 à 21:20, Brian E Carpenter a écrit :
> On 03-Nov-20 22:02, [e-mail] wrote:
>>> With 5G network slicing will become a formidable part of the 
>>> paradigm shift from shared resources to network resource 
>>> isolation at the radio and RAN layers.  The use case for being 
>>> able to provide critical segmentation for network slices makes
>>> it an industry imperative to be able support variable IID
>>> lengths with slaac as all 5G deployments worldwide will be IPv6
>>> only from the start.
>> 
>> You'd likely find most people on this group thinking of 5G as just 
>> "more of the same"; access to the Internet. At least I struggled 
>> understanding what there is about 5G that would require a change
>> in IPv6 addressing models.
>> 
>> Are there any of the problems/use-cases that cannot be solved by 
>> assigning a shorter prefix than /64 to the end-user networks?
> 
> Is it really the case that 5G hasn't fixed the notorious 4G bug of 
> only allowing /64s?

I think yes, it is really the case.

A 5G network will see many modifications (from 4G) with the New Radio
access, but is not likely to make many change to GGSN and SGSN, which 
are the entities involved in IP address allocation.  I think.

One operator's goal to migrate to 5G is to make it as easy to migrate
for end users as possible.  Just notice higher bandwidth.  On a
smartphone, there will be just a small acronym '5G', hardly seen without
glasses, in the list of acronyms GPRS, 3G, H+, 4G and 4G+ next to 5
small bars showing signal quality.  And no, 5 in the number of bars is 
just a coincidence: they have always been 5 even in times of 2G.

This is how it happened when it moved from 3G to 4G and still kept that /64.

Where I live the next 5G deployment for the masses arrives at the
earliest in December.  At that point it's easy to try it and answer to
the question you raise: is 5G offering something else than a /64 to
smartphones.

(there is one little issue in trying that very quickly, in that the
operator that claims to do 5G first among all the others also has
something against IPv6, believing in essence IPv6 is not needed
anywhere, including 5G; so the trial might not happen that quickly).

> Or is it a matter of operational choice by 5G operators?

I dont know.

Some operators say they depend on what the router manufacturers do.
Router manufacturers do put this DHCPv6-PD option in, but for
convenience, also keep the 64bit limit in most places.

Alex

> 
> Brian
>